Klf15 expression is higher in WAT compared with BAT and is suppressed by β-adrenergic signaling. To elucidate the context-specific, as well as the most physiologically potent, activities of KLF15 in adipose tissue, we surveyed the expression levels of Klf15 across 3 major types of adipose: visceral, subcutaneous, and intrascapular brown in WT mice. Strikingly, we found that the expression level of Klf15 is approximately 75% lower in brown adipose tissue (BAT) than in white (WAT) (Figure 1A). These results suggest physiological implications for these differences, including raising the possibility of a requirement to downregulate Klf15 levels for proper brown fat function. Brown fat responds to β-adrenergic signaling with the induction of thermogenesis (13), and this is a primary physiological function of this tissue. Therefore, we tested the impact of β-adrenergic stimulation on Klf15 expression in adipose tissues. We found that exposing WAT to the β-adrenergic stimulant isoproterenol in tissue culture resulted in downregulation of Klf15 expression (Figure 1B). To test if these results were relevant in vivo, we injected WT mice with the β-adrenergic agonist CL-316243 and discovered that this treatment results in an approximately 50% downregulation of Klf15 expression levels in WAT (Figure 1C). There are 3 different adrenergic receptor family members (ADRB 1–3). Interestingly, the 3 adrenergic receptors have distinct properties as well as both ligand-dependent and constitutive activation (reviewed in ref. 34). We compared the expression levels of these receptors across the adipose types and found that Adrb1 is the most differentially expressed in BAT compared with WAT (Figure 1D and Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/JCI172360DS1). An analogous pattern was observed when the expression levels of the adrenergic receptors were compared in human white and brown adipocytes (35). Adrb1 is a particularly interesting, and perhaps distinct, adrenergic receptor; for example, overexpression of Adrb1 in white adipocytes is sufficient to generate constitutive activity and mimics the effect of systemic infusion of agonists, even in the absence of additional agonists (36). These findings also indicate that increasing the expression level of the Adrb1 alters white adipocytes in contexts where other adrenergic receptors are present.

Acute modulation of Klf15 expression in white adipocytes induces a beige fat expression profile. (A) RT-qPCR quantifying the expression levels of Klf15 in adipocytes isolated from distinct fat pads. 1-way ANOVA, n = 3. (B) RT-qPCR quantifying the expression levels of Klf15 in adipocytes after isoproterenol (Iso) treatment for 4 hours. Student’s t test, n = 6. (C) RT-qPCR quantifying the expression levels of Klf15 in iWAT isolated from mice after they were injected with CL316243 (CL) (1 mg/kg/day) for 7 days. Student’s t test, n = 4. (D) RT-qPCR quantifying the relative expression levels of Adrb1, Adrb2, and Adrb3 in iWAT, gWAT, and BAT with iWAT set to 1 for each litter. 1-way ANOVA, n = 5. (E) Light phase microscopy images of adipocytes from differentiated SVF harvested from the iWAT of Klf15-fl/fl mice and infected with adenovirus expressing Cre (Ad-Cre) or adenovirus control (Ad-control). Scale bar: 25 μm. (F) RT-qPCR quantifying the expression levels of thermogenic genes and panadipocyte marker Adipoq. Student’s t test, n = 4. (G) RT-qPCR quantifying the expression levels. Student’s t tests followed by Holm-Šidák correction, n = 7. (H) Immunoblots detecting and quantifying the relative levels of β1AR following acute Klf15 deletion in adipocytes. Student’s t test, n = 3. (I) Immunoblots detecting and quantifying the relative levels of phosphorylation of p38 MAPK following acute Klf15 deletion in adipocytes. Student’s t test, n = 3. (J) RT-qPCR quantifying the change in Ucp1 expression levels with SB202190 pretreatment. Student’s t test, n = 3–4. (K) RT-qPCR quantifying the expression levels of Ucp1 following acute Klf15 deletion in adipocytes treated with Isoproterenol. 2-way ANOVA, n = 3. *P < 0.05, **P < 0.01, ***P < 0.001.

Deletion of Klf15 disrupts maintenance of white adipocyte properties and induces Adrb1 expression. To begin to probe the physiological relevance of the above findings, we generated a Klf15-floxed mouse line using CRISPR/Cas9 to insert loxP sites surrounding exons 1–3 of the Klf15 gene (Klf15-floxed). We then harvested white adipocytes from these mice and infected them with an adenovirus that expresses Cre recombinase (cleaving the loxP sites, knocking out Klf15) or control adenovirus (Figure 1E). Cre-infected adipocytes had lower expression levels of Klf15 compared with controls (Figure 1F), demonstrating that Cre exposure properly results in efficient deletion of Klf15 in the cells. We discovered that deletion of Klf15 substantially induces the expression of genes that are critical to brown fat identity and function, including the expression of the canonical brown fat gene Ucp1 (Figure 1F). These results are particularly striking because they support that there is plasticity in mature adipocytes, where core properties of white adipocytes can be modulated independently from adipogenesis.

When we measured the expression levels of β-adrenergic receptors in response to deletion of Klf15, we discovered that Adrb1 was upregulated, while the other adrenergic receptors were unaffected or down regulated (Figure 1G). β1AR protein levels also increase with Klf15 deletion (Figure 1H). These findings are intriguing, in part, because overexpression of Adrb1 in subcutaneous white adipose tissue (iWAT) is sufficient to induce Ucp1 expression and beiging of this white fat depot (36, 37). Further, β1AR is more sensitive to stimuli than other adrenergic receptors (38). Therefore, our results suggest a plausible pathway by which KLF15 could modulate the maintenance of white adipocytes and beiging in iWAT.

We next probed if the induction of brown fat genes that occurs after deletion of Klf15 is mechanistically connected to the increased expression of Adrb1. Adrenergic receptor activation leads to increased phosphorylation of p38 MAP kinase which, in turn, induces Ucp1 expression (39). To test if this pathway is stimulated by the deletion of Klf15, we monitored the levels of p38 phosphorylation. Indeed, deletion of Klf15 leads to a marked increase in the levels of phosphorylated p38 in mature adipocytes (Figure 1I). Further, we found that inhibition of p38 MAP kinase suppresses the induction of Ucp1 with Klf15 deletion (Figure 1J).

As mentioned, β1AR activation can occur by both ligand-dependent and constitutive mechanisms. As no additional ligand was added to observe the induction of brown fat genes with Klf15 deletion, our results raise the possibility that ligand-independent signaling was occurring because of the constitutive activation of the β1AR receptor. However, adrenergic receptor agonists are present in the cell culture medium (40), which could stimulate the pathway. Therefore, to integrate this question further, we assessed the effect of an adrenergic receptor antagonist. We discovered that the antagonist propranolol diminishes, but does not abolish, the induction of Ucp1 expression in Klf15-deleted adipocytes (Supplemental Figure 1B). Reciprocally, we tested the effect of stimulation with an adrenergic receptor agonist. These experiments revealed that the addition of an agonist enhances the induction of Ucp1 in Klf15-deleted adipocytes (Figure 1K). Together, these results support a model where the induction of Adrb1 that occurs with Klf15 deletion results in both ligand-dependent and constitutive β1AR activation in the mature adipocytes.

To investigate this activity of KLF15 selectively in mature adipocytes in vivo, we crossed the Klf15-floxed mice with transgenic mice expressing Cre recombinase under the control of the adiponectin promoter and regulatory elements (Adipoq-Cre). Adipoq-Cre mice have been verified to selectively express Cre in mature adipocytes (41), enabling conditional deletion of Klf15 in mice that have both floxed Klf15 and the Adipoq-Cre transgene (Adipo-Klf15–cKO mice) in their genomes. Indeed, we verified that Klf15 is efficiently and selectively knocked-out in mature adipocytes and not in other tissues (Figure 2A) nor in the adipose stromal vascular fraction (SVF) (Figure 2B) of Adipo-Klf15–cKO mice. When we harvested the WAT from Adipo-Klf15–cKO mice, we noted that the iWAT had a browner appearance than iWAT from either Adipoq-Cre or Klf15-floxed mice (Figure 2C). In addition, the mass of the iWAT was less in Adipo-Klf15–cKO mice compared with littermate controls (Figure 2D). We discovered that the iWAT from Adipo-Klf15–cKO had substantially higher expression levels of multiple genes involved in brown fat identity and function (Figure 2E) and that upregulation of Ucp1 was restricted to mature adipocytes isolated from the iWAT (Supplemental Figure 1C); we did not detect any induction of Ucp1 in either visceral WAT (gWAT) or BAT (Supplemental Figure 1, D and E).

Adipocyte-specific Klf15 KO promotes beige adipocyte formation in iWAT. (A) RT-qPCR quantifying the expression levels of Klf15 in iWAT, gWAT, BAT, and liver from WT and Adipo-Klf15 mice. Student’s t tests followed by Holm-Šidák correction, n = 3. (B) RT-qPCR quantifying the expression levels of Klf15 in the SVF and adipocyte fraction of iWAT from WT and Adipo-Klf15 mice. Student’s t test followed by Holm-Šidák correction, n = 3. (C) Representative images of in situ iWAT in WT and Adipo-Klf15 mice. Scale bar: 5 mm. (D) Quantification of iWAT mass as a percent of body weight in WT and Adipo-Klf15 littermates. Ratio paired t test, n = 5. (E) RT-qPCR quantifying the expression levels of thermogenic genes in iWAT of WT and Adipo-Klf15 littermates. Student’s t test, n = 5. (F–H) RT-qPCR quantifying the expression levels of adrenergic receptors in iWAT, gWAT, and BAT from WT and Adipo-Klf15 mice. Student’s t test followed by Holm-Šidák correction, n = 5. (I) RT-qPCR quantifying the expression levels of Adrb1 versus Adrb3 in iWAT in littermates of WT and Adipo-Klf15 mice. 1-tailed ratio paired t test, n = 4. (J) Immunoblots detecting the levels of β1AR protein in iWAT from WT and Adipo-Klf15 mice compared with the levels of β-actin controls. (K) Quantifying the relative protein level of β1AR. Student’s t test, n = 4. *P < 0.05, **P < 0.01, ***P < 0.001.

We also tested if there were analogous changes in the expression levels of the adrenergic receptors, as we discovered in our ex vivo experiments (Figure 1G). Consistent with our findings in tissue culture, the in vivo studies revealed that only Adrb1 was selectively upregulated in the iWAT of Adipo-Klf15–cKO mice (Figure 2, F–H). As multiple factors, including external ones, can affect adrenergic signaling, to rigorously verify these results, we directly compared the ratio of expression levels of Adrb1 to Adrb3 in Adipo-Klf15–cKO mice against sex-matched littermates as individual paired replicates. The results of these orthogonal testing of the findings, which controls for any ‘crosstalk’ between Adrb1 and Adrb3 (42), confirmed that Adrb1 expression in iWAT is selectively upregulated in Adipo-Klf15–cKO mice (Figure 2I and Supplemental Figure 1, F and G). Finally, we tested if the expression change translates into altered protein levels of β1AR. Using immunoblotting we established β1AR protein levels are increased in the iWAT of Adipo-Klf1–cKO mice (Figure 2, J and K).

Recently, noncanonical and adrenergic-independent thermogenic pathways have been described in adipocytes (43, 44). To evaluate if Klf15 affects these mechanisms, we measured the expression levels of genes involved in these pathways. We did not detect evidence that these pathways were altered by deletion of Klf15 (Supplemental Figure 1H). These results further support that modulation of Adrb1 by Klf15 is a primary mechanism for the beiging of mature adipocytes that we observed.

Together, our results indicate that deletion of Klf15 in the Adipo-Klf15–cKO mice disrupts maintenance of white adipocyte properties and induces beiging selectively in the mature adipocytes within the iWAT depot. To investigate this further, as well as test the findings using an independent approach, we generated a second mouse line by crossing Klf15-floxed mice with transgenic mice expressing Cre under the regulation of Prx1 promoter and regulatory element (Prx1-Cre). We (45) and others (46) previously demonstrated that this Prx1-Cre transgenic mouse line efficiently and preferentially targets the iWAT fat depot adipocyte progenitor cells over gWAT. Indeed, we found that mice generated by this cross that have both floxed Klf15 and the Prx1-Cre transgene (Prx1-Klf15 cKO) have substantially lower levels of Klf15 expression selectively in the iWAT compared with other tissues and with littermate controls (Figure 3, A and B). In addition, the iWAT depots specifically weigh less, are browner, and have smaller adipocytes than iWAT depots in control mice (Figure 3, C–E and Supplemental Figure 2, A and B). While total body weights were not significantly different (Supplemental Figure 2, C and D), this may have been affected by the observed adipose depot–restricted effects combined with variable compensatory changes in food consumption (Supplemental Figure 2E).

Enhanced beiging occurs in the iWAT of Prx1-Klf15–cKO mice. (A) RT-qPCR quantifying the expression levels of Klf15 in iWAT, gWAT, BAT, and liver from WT and Prx1-Klf15 mice. Student’s t tests followed by Holm-Šidák correction, n = 3. (B) RT-qPCR quantifying the expression levels of Klf15 in the SVF and adipocyte fractions of iWAT from WT and Prx1-Klf15 mice. Student’s t tests followed by Holm-Šidák correction, n = 3. (C) Quantification of iWAT mass as a percent of body weight in WT and Prx1-Klf15 mice. Student’s t test, n = 9. (D) Representative images of in situ iWAT from female WT and Prx1-Klf15 mice. Scale bar: 5 mm. (E) Representative images of H&E stained histological sections of iWAT from female WT and Prx1-Klf15 mice. Bottom images are magnified from the indicated square in the top images. Scale bar: 100 μm. (F–H) RT-qPCR quantifying the expression levels of (F) panadipocyte markers, n = 4 (G) white adipocyte markers, n = 3–8, and (H) thermogenic genes in iWAT, n = 6–11. Student’s t test followed by Holm-Šidák correction (I) RT-qPCR quantifying the expression levels of adrenergic receptors in iWAT from WT and Prx1-Klf15 mice. Student’s t test followed by Holm-Šidák correction, n = 4. (J) RT-qPCR quantifying the expression levels of Adrb1 versus Adrb3 in iWAT from littermates of WT and Prx1-Klf15 mice. 1-tailed ratio paired t test, n = 4. (K) Immunoblot detecting UCP1 and GAPDH protein levels in the iWAT of WT and Prx1-Klf15 mice. (L) Time course quantification of OCRs of iWAT isolated from WT and Prx1-Klf15 littermates after exposure to Xamoterol using a Seahorse Bioanalyzer. 2-way ANOVA, n = 5. (M and N) Whole-body heat generation (kcal) of mice undergoing cold exposure. (M) 2-way ANOVA, (N) ANCOVA, n = 5. *P < 0.05, **P < 0.01, ***P < 0.001.

Expression profiling revealed that the levels of panadipocyte-expressing genes were similar in Prx1-Klf15–cKO mice compared with littermate controls (Figure 3F). Notably, the Prx1-Klf15–cKO iWAT has decreased expression of some white adipocyte marker genes (Figure 3G) as well as significantly higher expression levels of multiple brown fat marker genes (Figure 3H). Further, the expression level of Ardb1, but not Ardb2 or Ardb3, is upregulated in Prx1-Klf15–cKO iWAT as quantified in absolute levels (Figure 3I) as well as relative to Adrb3 expression in littermate controls (Figure 3J). Immunoblotting confirmed that these expression differences translate into higher protein levels of UCP1 and β1AR in Prx1-Klf15–cKO iWAT (Figure 3K and Supplemental Figure 2F).

Having established, using 2 mouse model systems, that deletion of Klf15 in iWAT induces β1AR and disrupts mature white adipocyte maintenance, we next investigated the functional implications of these findings. We measured the energetics of primary iWAT harvested from Prx1-Klf15 cKO compared with littermate controls in response to the selective β1AR agonist Xamoterol as well as the adrenergic agonist Isoproterenol. We discovered that iWAT with Klf15 deletion had an enhanced response to agonist stimulation, resulting in substantially higher induced OCRs (Figure 3L and Supplemental Figure 2G).

The above results indicate that deletion of Klf15 stimulates upregulation of the β1AR and results in adipocytes with enhanced responsiveness to adrenergic stimulation. We next investigated the systemic implications of these findings in vivo. By monitoring singly housed mice in metabolic cages, we found that Prx1-Klf15 cKO mice have higher energy expenditure per total body mass compared with littermate controls, and expenditure becomes significantly further induced by acute cold exposure, a physiologically relevant stimuli (Figure 3, M and N). Of note, the Prx1-Klf15–cKO mice are also more competent at maintaining their body temperature during cold exposure, and this occurs without any change locomotor activity (Supplemental Figure 3, A and B). These findings demonstrate that, even though this activity of KLF15 is cell-autonomous and restricted to the iWAT depot in mice, deletion of Klf15 in iWAT has physiological implications.

Our ex vivo studies revealed that adipocytes with Klf15 deletion have increased Adrb1 expression and enhanced adrenergic sensitivity. We also found Adrb1 expression levels are increased in the iWAT of mice with Klf15 deletion. To assess the in vivo implications of these findings, we administered adrenergic agonists to the mice and monitored the effect on energy expenditure. We discovered that the Prx1-Klf15–cKO mice, housed in a thermoneutral environment, have enhanced energy expenditure immediately following injections with agonists, consistent with increased sensitivity to adrenergic stimulation during the short half-lives of these drugs (Supplemental Figure 3, C–F). As with cold exposure, the increased energy expenditure was not caused by increased activity (Supplemental Figure 3G).

KLF15 modulates adipocyte sensitivity to β-adrenergic stimulation. Having discovered that the deletion of Klf15 in iWAT has systemic implications in vivo, we were motivated to further define the mechanisms driving this effect. Our results indicate that the β1AR receptor is upregulated in iWAT in both of our mouse models with Klf15 deletions. Further, our findings demonstrate that this effect is cell-autonomous and occurs rapidly, as we detected marked changes in Adrb1 following acute deletion of Klf15 in cells (Figure 1G), suggesting that this change is proximal to, and potential directly caused by, KLF15 action. This model was particularly intriguing because adrenergic signaling stimulates a brown fat expression profile in iWAT and promotes cellular energy utilization in adipocytes through the induction of uncoupled respirations. Of note, prior work by others implicated β3AR as the primary adrenergic receptor expressed in white adipocytes in rodents (47). Our results raise the exciting possibility that targeting KLF15 could induce an alternative adrenergic input pathway for white adipocytes that promotes energy utilization. To test this model, we acutely deleted Klf15 in white adipocytes using adenoviral infection of adipocytes. We then measured the expression levels of Adrb1 and Ucp1 both at the basal state as well as after stimulation with Xamoterol. We discovered that not only are basal levels of Adrb1 and Ucp1 expression higher in adipocytes after acute deletion of Klf15 but also Klf15-deleted adipocytes have a more robust response to the β1AR agonist that is cell autonomous (Figure 4, A and B). Adrenergic signaling stimulates the induction of cAMP production within responsive cells (48). Therefore, we quantified cAMP levels in Klf15-deleted adipocytes. Indeed, we found that deletion of Klf15 resulted in increased cAMP levels in the adipocytes (Figure 4C). These findings indicate that deletion of Klf15 in white adipocytes enhances cellular sensitivity to adrenergic stimulation through the β1AR receptor.

KLF15 regulates the expression of Adrb1 in a pathway conserved in humans. (A and B) RT-qPCR quantifying the expression levels of Adrb1 and Ucp1 in adipocytes with or without Xamoterol treatment. Student’s t test, n = 5. (C) Quantifying of unstimulated intracellular cAMP levels in adipocytes with deletion of Klf15. Student’s t test, n = 5. (D) Conservation of a canonical KLF15 binding site sequence (red) identified in the mouse Adrb1 gene. (E) Quantification of dual-luciferase assays on adipocytes transfected with the Adrb1 promoter-driven firefly luciferase reporter construct and pCMV-Klf15 or control plasmid and normalized by Renilla bioluminescence, facilitated by cotransfected pRL-TK plasmid. Student’s t test, n = 3. (F) RT-qPCR quantifying the amount of immunoprecipitated DNA containing the putative KLF15 binding site located in Adrb1 using a KLF15 antibody in iWAT adipocytes from WT and Prx1-Klf15–cKO mice. Student’s t test, n = 3. (G) RT-qPCR quantifying the amount of immunoprecipitated DNA containing the putative KLF15 binding site located in the Adrb1 using the FLAG compared to IgG antibody in iWAT adipocytes isolated from Klf153xFLAG mice. Student’s t test, n = 3. (H) Image of PCR amplicons in an agarose gel of the putative KLF15 binding site (Target region) compared with amplification of the control region from the ChIP of adipocytes from WT and Prx1-Klf15–cKO mice with the KLF15 antibody. (I) RT-qPCR quantifying the expression levels of Ucp1 in iWAT from mice injected with saline, Denopamine (10 μg/g/day), Xamoterol (8 ng/g/day), or Dobutamine (10 μg/g/day) for 7 days. 1-way ANOVA, n = 4–5. (J) Light phase microscopy images of human adipocytes. Scale bar: 25 μm. (K) RT-qPCR quantifying the expression levels of hKLF15, hADRB1, and hUCP1 in human adipocytes infected by Ad-shCtrl or Ad-shKLF15. Student’s t test followed by Holm-Šidák correction, n = 4. (L) OCRs and (M) respiratory profile in Ad-shKLF15 and Ad-shCtrl infected human adipocytes quantified using a Seahorse Bioanalyzer, 2-way ANOVA n = 8. (N) Time-course of OCRs of human adipocytes after exposure to Xamoterol. 2-way ANOVA, n = 8–9. *P < 0.05, **P < 0.01, ***P < 0.001.

Adrb1 is a KLF15 target gene. KLF15 functions as a transcription factor to both positively and negatively regulate the expression of genes by binding to a canonical DNA binding site sequence (Supplemental Figure 4A) in target genes (49). The rapid and substantial induction of Adrb1 expression after Klf15 knockdown suggested that expression of Adrb1 could be directly modulated by KLF15. Interestingly, there is context-specificity to KLF15 action, as Klf15 expression was found to stimulate expression of the α-adrenergic receptor Adra1a in models of nephric tubules (50). To probe deeper into the specific relationship between Klf15 and Adrb1 in white adipocytes, we acutely overexpressed Klf15 in adipocytes using adenoviral infection; we discovered that overexpression of Klf15 downregulates Adrb1 expression (Supplemental Figure 4B), further suggesting that Adrb1 is directly regulated by KLF15. Therefore, we scanned the promoter proximal region of the Adrb1 gene and identified a KLF15 binding site sequence 23 bps downstream of the transcriptional start site of Adrb1. Importantly, we found that this putative binding site is highly conserved, including in humans (Figure 4D). We subcloned the endogenous Adrb1 promoter and proximal region, containing this putative KLF15 binding site, into a promoterless luciferase reporter construct. We then transfected adipocytes with this luciferase reporter and cotransfected with either a Klf15 expression vector or an empty control vector. Luciferase assays revealed that cotransfection with the Klf15 expression vector inhibited Adrb1 promoter activity compared with control infected cells (Figure 4E). Further, mutating the putative KLF15 biding site using site-directed mutagenesis of the Adrb1 luciferase reporter construct (Supplemental Figure 4C) prevented KLF15 inhibition of Adrb1 promoter activity (Supplemental Figure 4D), indicating that this binding site was necessary and sufficient for this KLF15 activity. Finally, we performed chromatin immunoprecipitation (ChIP) studies using primary iWAT isolated from WT mice as well as mice with a flag (3×) knocked into the Klf15 gene (51) (Klf153xFLAG) compared with iWAT from Klf15-cKO mice as a control. We performed ChIP on the putative KLF15 binding site in the Adrb1 gene and found enriched occupancy at that site in both the iWAT of Klf153xFLAG mice and in WT mice (performed with anti-KLF15 and anti-FLAG, respectively) as well as compared with a distinct site located adjacent to the KLF15 binding site (Figure 4, F–H). Together, these data indicate that this site is a bona fide KLF15 binding site and response element in the Adrb1 gene.

KLF15 activity is conserved in primary human adipose cells. We next tested for evidence that this pathway is conserved and relevant to humans. First, we injected WT mice with a panel of β1AR agonists, including ones that are FDA approved drugs for use in humans, and collected iWAT tissue. Strikingly, we discovered that these drugs provoked a substantial induction of Ucp1 expression in the iWAT (Figure 4I). Next, we purchased human primary SVFs harvested from subcutaneous abdominal white adipose tissue biopsies. We differentiated these cells in culture and then knocked down human KLF15 (hKLF15) expression in the mature human adipocytes using adenoviral infection of Klf15 shRNA compared with control shRNA infection (Figure 4J). This approach resulted in a greater-than 80% knockdown of hKLF15 in the human adipocytes (Figure 4K). Strikingly, acute knockdown of hKLF15 resulted in a marked induction of both hARDB1 and hUCP1 in the primary human adipocytes (Figure 4K). Importantly, we found that induction of hUCP1 was hADRB1 dependent (Supplemental Figure 4E), implicating this pathway as the primary mechanism for the observed induction of hUCP1. In further support of this result, while knockdown of hADRB1 blocked the induction of hUCP1, it did not alter the level of phosphorylated hormone-sensitive lipase (pHSL) (Supplemental Figure 4, F and G), indicating that the increased level of lipolysis with hKLF15 knockdown observed by us (Supplemental Figure 4H) and others (32) is not the major driver of the induction hUCP1 expression.

We next tested functional effects of hKLF15 in human adipocytes by measuring cellular energetics. Using a Seahorse analyzer, we discovered that knockdown of hKLF15 induces increased OCRs in human white adipocytes (Figure 4, L and M) without changing mitochondrial content (Supplemental Figure 4I); analogous to what we observed in our studies using the mouse models described above. Finally, we discovered that knockdown of hKLF15 in human white adipocytes resulted in an enhanced induction of the OCR in response to adrenergic agonists Xamoterol and Dobutamine (Figure 4N and Supplemental Figure 4J and K). These results support that targeting hKLF15 increases sensitivity to adrenergic stimulation in human adipocytes.